Load balancing for multi-robot construction

In distributed multi-robot construction it is important to set the relative rates at which different construction sites receive raw building materials. Otherwise, subtasks finish at different times introducing unnecessary delays. We present a feedback algorithm to achieve robust load balancing in routing building materials for stochastic, distributed, multi-robot construction systems. We express global behavior in terms of local reactive behavior via Guarded Command Programming with Rates and prove correctness of the load-balancing controller for a wide range of conditions. We adapt a proof from earlier work on controlling Stochastic Chemical Kinetic systems and illustrate the algorithm on the Factory-Floor robotic testbed [1].

Authors

• Nils Napp
• Eric Klavins

Keywords

• Raw materials
• Routing
• Robots
• Production facilities
• Markov processes
• Floors
• Robustness
• Load Balancing
• Building Materials
• Construction Sites
• Multi-agent Systems
• Local Behavior
• Markov Chain
• State Space
• Feedback Control
• Global Status
• Physical System
• Control Region
• Error Propagation
• Feed Rate
• System Scale
• Part Of The State
• Error Signal
• Generator Matrix
• End-effector
• Construction Program
• Factory Floor
• Discrete State Space
• Steady-state Distribution
• Composition Operator
• Building Programs
• Output Function
• Contributions Of This Paper
• Common Resources
• Fault-tolerant